本文通过一个案例，介绍优化器对单列索引和联合索引的选择。 order表的ord_seq字段上有2个索引，单列索引（ order_seq）和联合索引（order_seq,order_type）
MySQL > explain select * from `order` where order_seq = 1502131212577; +----+-------------+-------+------------+------+-------------------------------+------------+---------+-------+------+----------+-------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+-------+------------+------+-------------------------------+------------+---------+-------+------+----------+-------+ | 1 | SIMPLE | order | NULL | ref | idx_ordseq,idx_ordseq_ordtype | idx_ordseq | 8 | const | 1 | 100.00 | NULL | +----+-------------+-------+------------+------+-------------------------------+------------+---------+-------+------+----------+-------+
可以看到优化器选择了单列索引idx_ordseq( order_seq)，而不是联合索引。因为该索引的叶子节点包含单个键值，所以理论上一个页可以存放更多的记录。 如果换种场景：
MySQL > explain select * from `order` where order_seq = 1502131212577 order by order_type desc; +----+-------------+-------+------------+------+-------------------------------+--------------------+---------+-------+------+----------+-------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+-------+------------+------+-------------------------------+--------------------+---------+-------+------+----------+-------------+ | 1 | SIMPLE | order | NULL | ref | idx_ordseq,idx_ordseq_ordtype | idx_ordseq_ordtype | 8 | const | 1 | 100.00 | Using where | +----+-------------+-------+------------+------+-------------------------------+--------------------+---------+-------+------+----------+-------------+ 1 row in set, 1 warning (0.00 sec)
where条件字段和排序字段添加联合索引，解决了filesort的问题。
